KIRA

A smart bot trained using Ineuron twitter dataset.

Preprocessing the data :

We process the given data by extracting the text from json file , and pairing the questions with answers. In the end we got more 4 lakh question answer pair sets. We used this data for our prediction of answer to repetitive questions.

Processing the data :

We have used NLTK library to process our data .

• We dropped duplicated questions set.
• We dropped null questions set.
• We removed stopwords and lemmatized the question text.
• We tokenize the text.
• By using above steps we transformed the data.

User input :

• User will enter a question
• The question will be transformed by using above steps
• The tokenized question will be compared with saved data and mathcing answer will be printed out.
• The question which is similar to asked question will be printed out

Production

• Use Gunicorn gunicorn --bind 0.0.0.0:$PORT --reload wsgi:app